Strictly speaking even Tim's solution may not give the true 'smallest
box', by which I assume is meant 'minimal volume box'. I suspect
moleman does this calculation (my apologies to Gerard if I'm wrong!)
by determining the axes of inertia, then rotating the molecule by the
matrix of eigenvectors and finally determining the limits with respect
to this rotated frame of reference. This will work fine if the
protein is roughly ellipsoidal (including of course spherical as a
special case), but if it has an irregular shape, then it may well not
give the minimal volume box. To do this properly is not quite as
simple as it looks, it essentially requires a minimisation of the
volume w.r.t. the box parameters (length and orientation of the axes).
Probably a simplex method would be a good way to do this, starting
from the inertial axes solution as an initial approximation. So my
answer is that AFAIK there is no program that will do this.
Whether it's worth going to this degree of effort to write a program
of course depends greatly on the purpose to which the results will be
put. Maybe if you said why you need this, then we can assess how
accurate you need the result to be, and tailor the suggestions
accordingly.
Cheers
-- Ian
On Wed, Mar 24, 2010 at 10:22 AM, Tim Gruene <[log in to unmask]> wrote:
> Dear Francois,
>
> I don't know if the pdbset command Clemens and Eleanor suggested actually
> gives you the answer you are looking. Take for example this case:
>
> ---
> | /|
> |/ |
> ---
>
> assume the diagonal line is the protein as it is inside the unit cell, the
> unit cell dimensions are not the smallest box that contains you diagonal line.
> To get the smallest box you have to rotate the line by 45degrees and then
> re-measure.
>
> To get this answer I would use moleman2 :
> > read your.pdb
> > xyz align
> > stats
> Then that statistics output would give you the mostly correct answer.
>
> As I said, I don't want to blame pdbset, because I simply don't know whether it
> takes this into account.
>
> Cheers, Tim
>
> On Wed, Mar 24, 2010 at 04:18:04PM +0900, Francois Berenger wrote:
>> Hello,
>>
>> Is there a ccp4 tool to find automatically the smallest virtual
>> orthogonal "box" that contain a given PDB ?
>>
>> Even if your favorite tool is not part of ccp4, I would be happy
>> to know about it. ;)
>>
>> Thanks,
>> Francois.
>
> --
> --
> Tim Gruene
> Institut fuer anorganische Chemie
> Tammannstr. 4
> D-37077 Goettingen
>
> GPG Key ID = A46BEE1A
>
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
>
> iD8DBQFLqefrUxlJ7aRr7hoRAnpYAJ4tj3x96F0DAU4fC8W3P525rajaGwCeJxq7
> 17/s/x4/dQlc1Fh1luo2iPw=
> =X9eu
> -----END PGP SIGNATURE-----
>
>
|